Google Cloud Instance Schedules: Auto Start/Stop VMs and Save 40%

The easiest cloud cost win you are leaving on the table

The biggest cost in any virtual desktop or dev environment is compute that runs when nobody uses it. If your office works 9–6, Monday to Saturday, why are your VMs billed 24×7?

Google Cloud's native Instance Schedules fix this with a start time and a stop time attached to each VM.

What Instance Schedules do

• Start time: VM powers on automatically (e.g., 8:45 AM)
• Stop time: VM shuts down automatically (e.g., 6:15 PM)
• Runs on a defined weekly schedule (Mon–Sat, or any pattern)
• Fully native to GCP — no cron server, no scripts, no extra cost

How to set one up

1. In Cloud Console, go to Compute Engine → Instance Schedules
2. Create a schedule with start/stop times and days
3. Attach it to the VMs you want on that schedule
4. Done — GCP handles the rest

Limits to know (2026)

• One schedule per VM (plan your tiers carefully)
• A schedule can be attached to up to 1,000 VMs
• Schedule and VM must be in the same region

Real result: servers that sleep

For a 100+ employee real estate company we classified every server into three tiers:

| Tier | Schedule | Example | |

|

|

| | Always-on | 24×7 | Domain controllers, RD Gateway, file servers | | Business-hours | 9–6 Mon–Sat (Instance Schedules) | Office RDS session hosts, app servers | | On-demand | Started on request, 15-min idle shutdown | CAD session hosts |

The compute bill dropped by ~40% — simply because servers now sleep when nobody needs them.

The 15-minute idle watchdog

GCP's own idle detection is a recommendation-based check over 1–14 days — too slow for on-demand pools. So for CAD hosts we added a custom 15-minute in-guest idle watchdog with Cloud Monitoring as a safety net. Last user logs out → server stops within 15 minutes.

📘 Read the full case study: Servers That Sleep — 100-Seat RDS on GCP

Want your cloud bill cut? WhatsApp +91 93197 07938 or contact@krizia.in.